Noise filtering and vibration suppression method and system for scanning micromirror

Abstract

The invention relates to a noise filtering and vibration suppression method and system for a scanning micromirror, and the method comprises the steps: constructing a motion characteristic model of the scanning micromirror without vibration, and obtaining the model output of the scanning micromirror without vibration; modeling the acquired multi-source fusion noise and unknown vibration interference of the scanning micromirror into a Gaussian mixture model and vibration impact interference; gaussian mixed noise parameters are determined, and a single filtering updating part is converted into interactive multi-model fusion updating; on the basis of interactive multi-model fusion updating, constructing an updated part into a vibration suppression target function in a nonlinear regression form, and seeking to minimize the vibration suppression target function; and based on the filtering result of each part and the posterior covariance matrix, calculating an updated likelihood function and a model probability, and obtaining a fused filtering value, thereby realizing adaptive filtering and vibration suppression of the scanning micromirror. Compared with the prior art, the method has the advantages of effectively realizing adaptive filtering and vibration suppression without depending on an additional vibration measurement sensor and the like.

Description

Technical Field

[0001] The present invention relates to the technical field of scanning micromirror measurement data processing, and in particular to a noise filtering and vibration suppression method and system for a scanning micromirror. Background Art

[0002] Automotive LiDAR based on scanning micromirrors features high resolution, long detection range, and good coherence, and is commonly used in fields such as autonomous driving, robotics, and environmental monitoring. Due to the diverse application scenarios, complex deployment environments, and high measurement accuracy requirements, ensuring the accuracy and stability of LiDAR signals requires reliable closed-loop control of the scanning micromirrors, as well as adaptive filtering and vibration suppression of the measurement signals.

[0003] On the one hand, the application scenarios of lidar determine that the noise environment it faces is quite complex, and general adaptive filtering methods cannot achieve high-precision filtering; on the other hand, the vibrations in industrial sites and vehicle-mounted environments are random, have a wide spectrum distribution, and are difficult to measure directly. If an independent vibration measurement sensor is used, it will not only be large in size and have limited spectral adaptability, but also increase the cost of the overall lidar solution.

[0004] After searching, Chinese invention patent application publication number CN118171014A discloses a method and system for suppressing the effects of unknown vibration on a visual precision measurement device. The method comprises the following steps: constructing a vibration-free motion characteristic model of the visual precision measurement device and obtaining the output of the vibration-free motion characteristic model; obtaining the motion output displacement of the visual precision measurement device affected by the unknown vibration; estimating the unknown vibration based on the error between the output of the vibration-free motion characteristic model and the motion output displacement of the precision measurement device affected by vibration; constructing an anti-vibration objective function based on the estimated unknown vibration, and seeking an optimal unknown vibration compensation amount to minimize the anti-vibration objective function; and compensating the visual precision measurement device based on the optimal unknown vibration compensation amount to complete the suppression process. This existing patent application suffers from the problem of not suppressing noise.

[0005] How to achieve noise filtering and vibration suppression of scanning micromirrors at low cost has become a technical problem that needs to be solved. Summary of the Invention

[0006] The purpose of the present invention is to overcome the defects of the above-mentioned prior art and provide a noise filtering and vibration suppression method and system for a scanning micromirror.

[0007] The purpose of the present invention can be achieved by the following technical solutions:

[0008] According to one aspect of the present invention, a method for noise filtering and vibration suppression of a scanning micromirror is provided, the method comprising:

[0009] Construct a motion characteristic model of the scanning micromirror when there is no vibration, obtain the model output of the scanning micromirror when there is no vibration, and obtain filter-related parameters;

[0010] The multi-source fusion noise and unknown vibration interference of the scanning micromirror are modeled as a Gaussian mixture model and vibration impact interference;

[0011] Determine the Gaussian mixture noise parameters of the Gaussian mixture model to be modeled, and transform the single filtering update part into an interactive multi-model fusion update;

[0012] Based on interactive multi-model fusion updating, the updating part is constructed as a vibration suppression objective function in the form of nonlinear regression, and the minimization of the vibration suppression objective function is sought;

[0013] Based on the filtering results of each part and the posterior covariance matrix, the updated likelihood function and model probability are calculated, and the fused filtering value is obtained to achieve adaptive filtering and vibration suppression of the scanning micromirror.

[0014] Preferably, the motion characteristic model of the scanning micromirror when there is no vibration is specifically:

[0015]

[0016] y k =-θ1y k-1 -θ2y k-2 +θ3c k-1

[0017]

[0018] Where y k 、y k-1 and y k-2 The model outputs the mirror deflection angle when the angular motion characteristics of the scanning micromirror are k, k-2 and k-2 respectively without vibration influence, and c k and c k-1 They represent the unmeasurable electromagnetic torque related variables inside the scanning micromirror at time k and time k-1, respectively. k-1 is the control voltage signal of the scanning micromirror at time k-1, Δu k =u k -u k-1 is the system input difference, sgn(·) is the sign function, θ1, θ2, θ3, ξ, ρ, τ and are all parameters to be identified in the model.

[0019] Preferably, the multi-source fusion noise and unknown vibration interference are modeled as a Gaussian mixture model and vibration impact interference, specifically:

[0020]

[0021] Where, v k is the system measurement noise; ~ indicates that it follows the latter distribution; Indicates that the mean is μ i , the variance is Gaussian distribution; is the parameter to be estimated of the Gaussian mixture model, θ is an arbitrary constant satisfying 0<θ<<1, and K is the number of Gaussian components used to model multi-source fusion noise; The covariance is S k Unknown vibration disturbance; ∈ i 、 and These are all Gaussian mixture model parameters of multi-source fusion noise, which serve as relevant parameters of the adaptive filter.

[0022] Preferably, based on the determined Gaussian mixture model parameters, the single filtering update part is transformed into an interactive multi-model fusion update, specifically: according to the Gaussian mixture model parameters, the model transition probability is determined, and the fusion state of each model is estimated based on the fusion state of the last recursion. Calculate the fusion covariance matrix of each model based on the covariance matrix of the last recursion:

[0023]

[0024] in, is the estimated state of the i-th model at time k-1, In fusion state, is the covariance matrix of the i-th model at time k-1, is the fusion covariance matrix, and i is the model index.

[0025] Preferably, the vibration suppression objective function of constructing the update part into a nonlinear regression form is specifically:

[0026]

[0027] in, is the total cost, n+m is the vector e k The number of elements, n is the dimension of the state vector x, m is the dimension of the output vector y, is the log-cosine hyperbolic robust cost function.

[0028] More preferably, the method further comprises solving the vibration suppression objective function Vibration suppression of the scanning micromirror is obtained.

[0029] Preferably, the calculation process of the fusion filter value includes:

[0030]

[0031] Where i represents the index of the i-th interactive multi-model, K is the number of indicators of the interactive multi-model, is the interactive multi-model probability, is the estimated state of the i-th model at time k, is the fusion filter value.

[0032] According to another aspect of the present invention, a noise filtering and vibration suppression system for a scanning micromirror is provided, the system comprising:

[0033] Motion characteristic model module: used to construct the motion characteristic model of the laser radar micromirror device when there is no vibration and obtain the output of the motion characteristic model when there is no vibration;

[0034] Sensor output module: used to obtain the output generated by the laser radar scanning micromirror due to multi-source fusion noise and vibration interference;

[0035] Multi-source fusion noise estimation module: This module estimates the multi-source fusion noise parameters based on the output of the motion characteristic model in the absence of vibration and the error sliding window between the outputs affected by the multi-source fusion noise and vibration interference.

[0036] Vibration suppression module: based on the vibration suppression objective function, seeks to minimize the vibration suppression objective function;

[0037] Fusion filtering module: Based on the output of the vibration suppression module and the multi-source fusion noise parameters estimated by the multi-source fusion noise estimation module, the fusion filtering value is calculated to achieve adaptive filtering and vibration suppression of the scanning micromirror.

[0038] Preferably, the multi-source fusion noise estimation module includes a lidar micromirror noise estimation submodule and a fusion filter parameter estimation submodule, wherein the lidar micromirror noise estimation submodule is used to determine the parameter estimation of the Gaussian mixture model of the multi-source fusion noise estimation module based on the motion characteristic model module and the sensor output module.

[0039] More preferably, the fusion filter parameter estimation submodule is used to determine the parameters required by the fusion filter module according to the interactive multi-model update.

[0040] Compared with the prior art, the present invention has the following beneficial effects:

[0041] (1) The present invention models multi-source fusion noise and unknown vibration interference through a sliding window of the difference between the kinematic model in the absence of vibration and the measurement value affected by the multi-source fusion noise and vibration, constructs a vibration suppression objective function through the Gaussian mixed noise parameter, seeks to minimize the vibration suppression objective function, obtains the fusion filter value after vibration suppression, and adaptively eliminates noise and suppresses vibration. Therefore, the present invention does not need to rely on additional vibration measurement sensors to effectively realize adaptive filtering and vibration suppression of the vehicle-mounted laser radar scanning micromirror.

[0042] (2) The present invention does not require the use of additional vibration measurement sensors, which not only avoids the problem of limited spectrum adaptability of commonly used vibration sensors, but also reduces suppression costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 Schematic diagram of the flow of the filtering and vibration suppression method of the scanning micromirror in the present invention;

[0044] Figure 2 Schematic diagram of the structure of the filtering and vibration suppression system of the scanning micromirror in the present invention;

[0045] Figure 3 This is a measurement output waveform diagram of the vehicle-mounted laser radar scanning micromirror in the present invention;

[0046] Figure 4 for Figure 3 The waveform diagram of the measured output after passing through the filtering and vibration suppression device of the present invention. DETAILED DESCRIPTION

[0047] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts should fall within the scope of protection of the present invention.

[0048] Example 1

[0049] This embodiment relates to a method for noise filtering and vibration suppression of a scanning micromirror of a vehicle-mounted laser radar. Figure 1 , the method comprises the following steps:

[0050] 101. Construct a motion characteristic model of the laser radar scanning micromirror when there is no vibration, obtain the model output of the scanning micromirror when there is no vibration, and obtain filter-related parameters;

[0051] 102. Obtain the multi-source fusion noise and unknown vibration interference faced by the scanning micromirror in actual working conditions, and model the multi-source fusion noise and unknown vibration interference of the scanning micromirror as a Gaussian mixture model and vibration impact interference; the multi-source fusion noise includes electrical noise, thermal noise, flicker noise, etc.

[0052] 103. Based on the constructed two-step noise processing framework, 1) in the prior part, multiple measurement values of the scanning micromirror are obtained in the form of a sliding window to determine the Gaussian mixture noise parameters of the Gaussian mixture model to be modeled; 2) based on the obtained Gaussian mixture model parameters, the single filtering update part is transformed into an interactive multi-model fusion update;

[0053] 104. Based on interactive multi-model fusion update, the logarithmic hyperbolic cosine robust cost function is introduced, the update part is constructed as a vibration suppression objective function in the form of nonlinear regression, and the vibration suppression objective function is minimized;

[0054] 105. Based on the filtering results of each part and the posterior covariance matrix, the updated likelihood function and model probability are calculated, and the fused filtering value is obtained to achieve adaptive filtering and vibration suppression of the scanning micromirror.

[0055] In 101, the motion characteristic model of the scanning micromirror without vibration can be constructed in the following way:

[0056]

[0057] y k =-θ1y k-1 -θ2y k-2 +θ3c k-1 (1)

[0058] Where y k 、y k-1 and y k-2 The model outputs the mirror deflection angle when the angular motion characteristics of the scanning micromirror are k, k-2 and k-2 respectively without vibration influence, and c k and c k-1 They represent the unmeasurable electromagnetic torque related variables inside the scanning micromirror at time k and time k-1, u k-1 is the control voltage signal of the scanning micromirror at time k-1, Δu k =u k -u k-1 , θ1, θ2, θ3, ξ, ρ, τ and are all parameters to be identified in the model.

[0059] In 102, the state and output angle of the scanning micromirror when it is disturbed by multi-source fusion noise and unknown vibration are affected by the system measurement noise v k Influence.

[0060] Multi-source fusion noise and unknown vibration interference are modeled as Gaussian mixture model and vibration impact interference, specifically:

[0061]

[0062] Where, v k is a random variable (i.e., system measurement noise); ~ indicates that it obeys the latter distribution; Indicates the mean The variance is Gaussian distribution; is the parameter to be estimated of the Gaussian mixture model, and θ is an arbitrary constant satisfying 0<θ<<1; The covariance is S k Unknown vibration disturbance. ∈ i 、 and These are the Gaussian mixture model parameters of multi-source fusion noise, which are derived by the expectation maximization algorithm to obtain the relevant parameters of the adaptive filter and the parameters needed later:

[0063]

[0064] in, is the model transition probability of the interactive multi-model, is the modified interactive multi-model probability, is the interactive multi-model probability, is the sum required for regularization.

[0065] In 103, based on the obtained Gaussian mixture model parameters, the single lidar micromirror filter update part is transformed into an interactive multi-model fusion update. Specifically, according to the Gaussian mixture model parameters of the noise, the model transition probability is determined by the following formula, and the fusion state of each model is determined according to the fusion state estimate of the last recursion According to the covariance matrix of the last recursion, the fusion covariance matrix of each model is calculated by the following formula:

[0066]

[0067] in, is the estimated state of the i-th model at time k-1, In fusion state, is the covariance matrix of the i-th model at time k-1, is the fusion covariance matrix, and i is the model index.

[0068] In 104, in order to suppress unknown vibration interference, the log-cosine hyperbolic robust cost function is introduced where e k for is an intermediate variable in the calculation process.

[0069] Influence function The nonlinear regression form of vibration suppression based on scanning micromirror is obtained:

[0070]

[0071] Among them, P k∣k-1 is the prediction covariance matrix at time k, P k-1∣k-1 is the posterior covariance matrix at time k-1, is the adaptive filter system matrix at time k, Q k is the covariance matrix of process noise; in and P k∣k-1 and R k The Cholesky decomposition of is the state estimation prediction value at time k-1, which is directly derived from formula (1); R k is the variance matrix of the measurement noise, D k In the above formula, there is (intermediate variable in the calculation process), To calculate the expectation, t k (x k ) is the intermediate variable required for calculation, e k To calculate the intermediate variables required, v k To measure noise, is the prior prediction state at time k, x k is the real state of the system, is the Cholesky decomposition, h(x k ) is the output function, y k is the micromirror output angle.

[0072] Vibration suppression objective function of scanning micromirror The expression is:

[0073]

[0074] in, is the total cost, n+m is the vector e k The number of elements, is the log-cosine hyperbolic robust cost function, e k,i is the vector e k The i-th element in .

[0075] To further solve the objective function The objective function is x k The partial derivatives of are as follows:

[0076]

[0077] The vibration suppression calculation process of the scanning micromirror is obtained:

[0078]

[0079] in, and are the ζth iteration value and the ζ+1th iteration value respectively, is the predicted value at time k, is K,y of the ξth iteration k is the output angle of the micromirror, h(·) is the output function, H(·) is the corresponding linear Jacobian form, is the damping coefficient, is the reweighted covariance matrix, The reweighted measurement noise variance matrix, is the reweighted prediction covariance matrix, I is the identity matrix, is the e of the ζth iteration k , and P k∣k-1 and R k Cholesky decomposition of .

[0080] Finally, the fusion filter value is calculated by the following expression:

[0081]

[0082] Where i represents the index of the i-th interactive multi-model, K is the number of indicators of the interactive multi-model, is the model probability, is the estimated state of the i-th model at time k, is the fusion filter value.

[0083] Example 2

[0084] This embodiment also relates to a method for noise filtering and vibration suppression of a laser radar micromirror, comprising:

[0085] Step 1: Construct a motion characteristic model of the laser radar micromirror when there is no vibration, and obtain the model output of the laser radar micromirror when there is no vibration.

[0086] The motion characteristic model of the LiDAR micromirror without vibration can be described as follows:

[0087]

[0088] Where y kis the mirror deflection angle output by the model describing the angular motion characteristics of the scanning micromirror when there is no vibration influence at time k, f is the system model for constructing the angular motion characteristics of the scanning micromirror measurement device when there is no vibration influence, and x k is the system state of the scanning micromirror at time k when there is no vibration influence, which is a third-order vector, where the state variable x 1,k is the deflection angle, the state variable x 2,k is the deflection angular velocity, the state variable x 3,k is an augmented variable used to represent variables related to the electromagnetic torque that cannot be measured inside the scanning micromirror, u k-1 is the system input of the scanning micromirror at time k-1, and h is the output model describing the angular motion characteristics of the scanning micromirror measurement device when there is no vibration influence.

[0089] The above model can be decomposed into formula (9). In order to proceed with the subsequent steps, it is necessary to solve the corresponding linear Jacobian form of formula (9):

[0090]

[0091] Where, There is a relationship between non-smooth and multi-valued mappings, which is derived using the generalized gradient method:

[0092]

[0093]

[0094] Among them, u k and u k-1 are the system inputs of the scanning micromirror at time k and time k-1, Δu k =u k -u k-1 , sup is the supremum, and inf is the infimum.

[0095] Through the above analysis, the motion characteristic model output of the laser radar micromirror in the absence of vibration and the relevant parameters required for subsequent steps can be obtained.

[0096] Step 2: Obtain the measurement output of the LiDAR micromirror when it is subjected to multi-source fusion noise such as electrical noise, thermal noise, and flicker noise, as well as unknown vibration interference, and model the multi-source fusion noise of the LiDAR micromirror as a Gaussian mixture model and strong vibration interference;

[0097] The laser radar micromirror is modeled as the multi-source fusion noise and unknown vibration interference as shown in expression (2), where: is the parameter to be estimated of the Gaussian mixture model, where 0<θ<<1, The covariance is S k Arbitrary unknown vibration disturbance.

[0098] Step 3: Based on the constructed two-step noise processing framework, the LiDAR micromirror measurement values at multiple moments are obtained in the form of a sliding window in the prior part to determine the Gaussian mixture noise parameters to be modeled;

[0099] The sliding window is used to store the laser radar micromirror's multiple moment measurements. Furthermore, ∈ i , and The specific parameters of the Gaussian mixture model of multi-source fusion noise of the lidar micromirror can be derived by the expectation maximization algorithm.

[0100] The observation equation in step 2 can be expressed by the i-th model in the interactive multi-model as follows:

[0101]

[0102] Where, In the prior part of the filter, in order to make the update part perform interactive multi-model fusion update, it is necessary to first calculate the transition probability from the i-th model to the j-th model. According to the transition probability, the fusion state of the prior part can be calculated. and the fusion covariance matrix of the j-th model At the same time, you can get in is the prior covariance matrix at time k, A k is the corresponding linear Jacobian form, Q k is the covariance matrix of the process noise.

[0103] Step 4: Based on the obtained Gaussian mixture model parameters, the single filter update part is transformed into an interactive multi-model fusion update; based on the interactive multi-model fusion update, a logarithmic hyperbolic cosine robust cost function is introduced, and the update part is constructed as a vibration suppression objective function in the form of nonlinear regression, and the vibration suppression objective function is minimized;

[0104] Based on interactive multi-model fusion update, a log-cosine hyperbolic robust cost function is introduced to suppress unknown vibration interference. Through the corresponding partial derivative influence function The nonlinear regression form of the scanning micromirror vibration suppression is obtained as Expression (3).

[0105] The objective function expression of scanning micromirror vibration suppression is (4), and the objective function is x k The partial derivative of can be calculated using expression (5), Denote as the corresponding weight function, the weight function matrix is

[0106] in:

[0107] Ψ x (e k )=diag[ψ(e k,1 ),ψ(e k,2 ),…,ψ(e k,n )],

[0108] Ψ y (e k )=diag[ψ(e k,n+1 ),ψ(e k,2 ),…,ψ(e k,m+n )]

[0109] Therefore, the objective function The solution of is obtained by expression (6).

[0110] In step 5, based on the filtering results and the posterior covariance matrix of each part, the updated likelihood function and model probability are calculated, and the fused filtering value is obtained to realize the adaptive filtering and vibration suppression of the lidar micromirror.

[0111] The posterior covariance matrix expression is:

[0112]

[0113] Among them, I n is the identity matrix, K k is the gain matrix, H(x k ) is the corresponding linear Jacobian form of the output function.

[0114] The likelihood function expression is:

[0115]

[0116] in, is the intermediate variable of the calculation, is the updated fusion filter value, is the sum required for regularization, is the intermediate variable, y k The model outputs the mirror deflection angle that describes the angular motion characteristics of the scanning micromirror without vibration at time k. is the measurement noise variance matrix of the jth interactive multi-model.

[0117] Based on multi-model probability The fusion filter value can be finally calculated:

[0118]

[0119] Where i represents the index of the i-th interactive multi-model.

[0120] Example 3

[0121] This embodiment also relates to a noise filtering and vibration suppression system for a laser radar micromirror, such as Figure 2 , including:

[0122] Motion characteristic model module 201: used to construct a motion characteristic model of the laser radar micromirror device when there is no vibration, and obtain the output of the motion characteristic model when there is no vibration;

[0123] Sensor output module 202: used to obtain the output generated by the laser radar scanning micromirror due to multi-source fusion noise and vibration interference;

[0124] Multi-source fusion noise estimation module 203: estimates multi-source fusion noise parameters based on an error sliding window between the output of the motion characteristic model when there is no vibration and the output caused by the multi-source fusion noise and vibration interference;

[0125] The vibration suppression module 204 is based on a vibration suppression objective function and seeks to minimize the vibration suppression objective function.

[0126] Fusion filter module 205: determines the fusion filter value after vibration suppression based on the output of the vibration suppression module and the multi-source fusion noise parameters obtained by the multi-source fusion noise estimation module.

[0127] The multi-source fusion noise estimation module 203 includes a lidar micromirror noise estimation submodule and a fusion filter parameter estimation submodule, wherein:

[0128] The LiDAR micromirror noise estimation submodule is used to determine the parameter estimation of the Gaussian mixture model of the multi-source fusion noise estimation module based on the motion characteristic model module and the sensor output module:

[0129] The fusion filter parameter estimation submodule is used to determine the parameters required by the fusion filter module based on the interactive multi-model update, where the parameters include ∈ i , mean and variance

[0130] like Figure 3 and Figure 4 , demonstrating that the implementation of the present invention achieves excellent adaptive filtering and vibration suppression for a LiDAR micromirror. This demonstrates that the present invention effectively implements adaptive filtering and vibration suppression without relying on additional vibration measurement sensors. This not only avoids the limited spectral adaptability of commonly used vibration sensors, but also reduces suppression costs.

[0131] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in the present invention, and such modifications or substitutions are intended to be within the scope of protection of the present invention. Therefore, the scope of protection of the present invention shall be subject to the scope of protection of the claims.

Claims

1. A method for noise filtering and vibration suppression of a scanning micromirror, characterized in that: The method includes: Construct a motion characteristic model of the scanning micromirror when there is no vibration, obtain the model output of the scanning micromirror when there is no vibration, and obtain filter-related parameters; The multi-source fusion noise and unknown vibration interference of the scanning micromirror are modeled as a Gaussian mixture model and vibration impact interference; Determine the Gaussian mixture noise parameters of the Gaussian mixture model to be modeled, and transform the single filtering update part into an interactive multi-model fusion update; Based on interactive multi-model fusion updating, the updating part is constructed as a vibration suppression objective function in the form of nonlinear regression, and the minimization of the vibration suppression objective function is sought; Based on the filtering results of each part and the posterior covariance matrix, the updated likelihood function and model probability are calculated, and the fused filtering value is obtained to achieve adaptive filtering and vibration suppression of the scanning micromirror.

2. The method for noise filtering and vibration suppression of a scanning micromirror according to claim 1, wherein: The motion characteristic model of the scanning micromirror when there is no vibration is specifically: y k =-θ1y k-1 -θ2y k-2 +θ3c k-1 Where y k 、y k-1 and y k-2 The model outputs the mirror deflection angle when the angular motion characteristics of the scanning micromirror are k, k-2 and k-2 respectively without vibration influence, and c k and c k-1 They represent the unmeasurable electromagnetic torque related variables inside the scanning micromirror at time k and time k-1, respectively. k-1 is the control voltage signal of the scanning micromirror at time k-1, Δu k =u k -u k-1 is the system input difference, sgn(·) is the sign function, θ1, θ2, θ3, ξ, ρ, τ and are all parameters to be identified in the model.

3. The method for noise filtering and vibration suppression of a scanning micromirror according to claim 1, wherein: Multi-source fusion noise and unknown vibration interference are modeled as Gaussian mixture model and vibration impact interference, specifically: Where, v k is the system measurement noise; ~ indicates that it follows the latter distribution; Indicates that the mean is μ i , the variance is Gaussian distribution; are the parameters to be estimated for the Gaussian mixture model, To satisfy An arbitrary constant, K is the number of Gaussian components used to model multi-source fusion noise; The covariance is S k Unknown vibration disturbance; ∈ i 、 and These are all Gaussian mixture model parameters of multi-source fusion noise, which serve as relevant parameters of the adaptive filter.

4. The method for noise filtering and vibration suppression of a scanning micromirror according to claim 1, wherein: Based on the determined Gaussian mixture model parameters, the single filter update part is transformed into an interactive multi-model fusion update. Specifically, according to the Gaussian mixture model parameters, the model transition probability is determined, and the fusion state of each model is estimated based on the fusion state of the last recursion. Calculate the fusion covariance matrix of each model based on the covariance matrix of the last recursion: in, is the estimated state of the i-th model at time k-1, In fusion state, is the covariance matrix of the i-th model at time k-1, is the fusion covariance matrix, and i is the model index.

5. The method for noise filtering and vibration suppression of a scanning micromirror according to claim 1, wherein: The vibration suppression objective function of constructing the update part into a nonlinear regression form is specifically: in, is the total cost, n+m is the vector e k The number of elements, n is the dimension of the state vector x, m is the dimension of the output vector y, is the log-cosine hyperbolic robust cost function.

6. The method for noise filtering and vibration suppression of a scanning micromirror according to claim 5, wherein: The method further includes solving a vibration suppression objective function Vibration suppression of the scanning micromirror is obtained.

7. The method for noise filtering and vibration suppression of a scanning micromirror according to claim 1, wherein: The calculation process of the fusion filter value includes: Where i represents the index of the i-th interactive multi-model, K is the number of indicators of the interactive multi-model, is the interactive multi-model probability, is the estimated state of the i-th model at time k, is the fusion filter value.

8. A system for executing the method for noise filtering and vibration suppression of a scanning micromirror according to any one of claims 1 to 7, characterized in that: The system includes: Motion characteristic model module: used to construct the motion characteristic model of the laser radar micromirror device when there is no vibration and obtain the output of the motion characteristic model when there is no vibration; Sensor output module: used to obtain the output generated by the laser radar scanning micromirror due to multi-source fusion noise and vibration interference; Multi-source fusion noise estimation module: This module estimates the multi-source fusion noise parameters based on the output of the motion characteristic model in the absence of vibration and the error sliding window between the outputs affected by the multi-source fusion noise and vibration interference. Vibration suppression module: based on the vibration suppression objective function, seeks to minimize the vibration suppression objective function; Fusion filtering module: Based on the output of the vibration suppression module and the multi-source fusion noise parameters estimated by the multi-source fusion noise estimation module, the fusion filtering value is calculated to achieve adaptive filtering and vibration suppression of the scanning micromirror.

9. The system according to claim 8, characterized in that The multi-source fusion noise estimation module includes a lidar micromirror noise estimation submodule and a fusion filter parameter estimation submodule, wherein the lidar micromirror noise estimation submodule is used to determine the parameter estimation of the Gaussian mixture model of the multi-source fusion noise estimation module based on the motion characteristic model module and the sensor output module.

10. The system according to claim 9, characterized in that The fusion filter parameter estimation submodule is used to determine the parameters required by the fusion filter module according to the interactive multi-model update.

Images